Residual Life Prediction Of Feed System Based On Hybrid Drive Digital Twin Model

As the "industrial mother machine" of the manufacturing industry,the intelligent development of CNC machine tools is related to national core competitiveness,industrial security,and economic security.In the process of intelligent development,establishing accurate and highly practical digital models is the foundation of intelligence.However,the current modeling methods have many shortcomings,especially in terms of model accuracy.In recent years,digital twin technology has gradually become a research hotspot.By building a twin model of large and complex physical equipment in the virtual space,it realizes the global mapping and real-time feedback of the virtual space to the physical world,and meets the content and goals of intelligent manufacturing.In addition,applying digital twin technology to predictive maintenance of CNC machine tools to predict the remaining service life can effectively solve the problem of high cost and long time consumption of traditional life prediction methods.This paper takes the CNC machine tool feed system as the research object,based on the concept and method of digital twin technology,and completes the overall design of the CNC machine tool feed system’s remaining service life,the construction and verification of the hybrid-driven digital twin model of the feed system,the adaptive updating of the digital twin model of the feed system,and the prediction of the remaining service life of the feed system based on the digital twin model.The specific work is as follows:Firstly,this article explores the relationship between faults and performance in the feed system of CNC machine tools,including the causes of faults,fault modes,and the impact of faults on performance.At the same time,the application of digital twin technology in the feed system of CNC machine tools is studied.Its structure and composition are explored,and its characteristics and functions are analyzed from the five dimensions of digital twin space,physical space,physical data,virtual and real interactive space,and twin data.The core technology and key application scenarios of digital twin models are explained.Based on this,in order to achieve health monitoring and predictive maintenance of the feed system of CNC machine tools,the feasibility of digital twin technology in life prediction is studied,and a solution for applying digital twin models to the residual life prediction of the feed system of CNC machine tools is proposed.Secondly,in order to build a multi-domain digital twin mechanism model,a detailed analysis was carried out on the internal operation mechanism of the feed system.The mechanism model was divided into three modules:electrical subsystem,control subsystem,and mechanical subsystem,and a multi-level hierarchical modeling method was designed.Simulink/Simscape,a multidomain modeling tool,was selected to complete the construction of the digital twin mechanism model.In addition,a modeling method for the digital twin data-driven model of the feed system based on long-short-term memory neural network(LSTM)was proposed.By collecting a large amount of experimental data and normalizing it,reasonable training parameters were selected as the input and output of the LSTM,and the precise construction of the digital twin data-driven model was realized,which was integrated with the mechanism model to jointly constitute the digital twin model of the feed system.To verify the accuracy of the digital twin model,a virtual-real synchronous motion experiment was conducted on a dual-axis feed system.The results showed that the digital twin model constructed by this method had an accuracy of over 95%,which can provide strong support for the intelligent maintenance and optimization of CNC machine tools.Then,a digital twin model updating framework is proposed,consisting of four parts:physical layer,perception layer,model updating layer,and digital twin body layer,which can achieve real-time updating and optimization of the digital twin model.On this basis,an online updating method for the digital twin model based on a recursive least squares algorithm with a forgetting factor is introduced.This method considers the importance of historical data and the weight of updating factors,and can quickly and accurately update the digital twin model.Through experiments on a dual-axis feed system under variable working conditions,it is verified that the online updating method of the digital twin model has high accuracy and practicality and can effectively improve the adaptability of the model.Finally,the mechanism of performance degradation of the positioning error in the feed system of CNC machine tools was analyzed,and an innovative method for predicting the positioning error of the feed system based on a digital twin model was proposed.The method uses digital twin model simulation to predict the positioning error under different clearances,and based on the distribution law of the performance degradation amount of the positioning error,a degradation amount distribution model and a threshold distribution model based on normal distribution were constructed.Furthermore,a remaining life prediction model was established to predict the remaining life of the CNC machine tool.